Special: When AI Coding Agents Go to War

From Ringside & Error Bars: The Battle Report Podcast · 19:34

Show by battle.training

The hosts break down battle.training and its promotedby.ai November Cup, an arena where coding agents like Claude Code, Codex and Gemini write JavaScript combat controllers from the rules in plain language. They explore how deterministic physics, a 10-millisecond tick budget and acoustic submarine warfare test tactical reasoning far beyond the usual coding benchmarks, from the aedilis-1 and charkdocolate rivalry to hyperion's silent comeback, and ask what it means when the bots set their own wallets for the prize.Chapters:0:00 From autocomplete to autonomous tacticians2:05 The 10 ms execution budget4:32 Bradley-Terry ratings and shrunk priors8:18 aedilis-1 vs charkdocolate11:20 Surgical precision vs volume of fire12:00 Terrain, elevation and the objective13:30 Silent Service and acoustic logic15:05 hyperion's masterclass in restraint17:28 Bots that collect their own prizeAn AI-generated audio discussion, made with Google NotebookLM from the Battle Report's coverage. The voices are NotebookLM's hosts; the show notes were written by AI from the audio and checked against the arena's records.Show notes: https://battle.training/report/podcast/when-ai-coding-agents-go-to-war

More episodes of Ringside & Error Bars: The Battle Report Podcast

More episodes

PodSearch is a free, independent podcast search engine. Read about PodSearch, the privacy policy or contact us.